Ask HN: please review my technology.
Thinking about the latest video streaming news, I thought about a feasible way to reduce the costs related to video distribution.
I wanted to use p2p to reduce bandwith while retaining a simple user experience, staying in the browser. The only feasible solution I saw was to use a Java applet, Although it wasn't maybe the best option from the user experience perspective, I have to admit the latest JVM update has reasonable startup time and I also saw some users that didn't even realize that it was java (that's good).
If you want to see the result, you can take a look at a few samples here: http://www.bitlet.org/video
Please note that data is downloaded via bittorrent, if you have it blocked by a firewall, you won't be able to see anything :°(
I'd really like to know what you think about the product, what you would change, any feedback, idea, or critic.
Thank you
26 comments
[ 2.3 ms ] story [ 99.9 ms ] threadAnd why use Theora? Most clips are MPEG-4, reencoding to theora will cause quality loss, so your video player will always be associated with poor quality.
You have a very good idea, but there are a few major flaws.
What applet are you using? Did you program it yourself?
I used theora because I wanted a free video codec.
The bittorrent client is coded by myself with help and support from a friend, the video codec is jheora from fluendo, it also uses jogg and jorbis from jcraft.
1) Being forced to click "Trust" just to watch a video is pretty bad, and would turn away many of your users.
2) Having an alert box "Are you sure you want to leave this page?" is also pretty bad, and unnecessary.
Overall, I can see the value that this technology gives to you (less bandwidth). But what value do you give to the consumer? Why would I want to endure the various hassles of this site, and give up some my scarce upstream bandwidth, when I could just go to Youtube or Vimeo instead?
1) You're right: It's bad, but in order to open connections to different clients it needs a signed applet. I also considered a custom plugin but I thought forcing a user to install it could be ever worse. Do you have any better idea?
2) True, will remove it.
I think reducing publisher's costs could lead to benefits even for consumers, I'm thinking about hd videos.
If you employed this technology & it took off I suspect you would be faced with very angry ISP's who may well ban access from their networks.
Plus of course it eats right into the users BW - they theoretically will upload more than they download and with limited BW allowances people could easily start gettign traffic shaped without even knowing why!
That said it's a neat solution to the BW problem.. something of a catch 22.
Ask any ISP :) P2P sucks for them.
And while I don't think ISPs really care about you maxing out a phone line, they will notice you maxing out a broadband line. And bittorent will do that.
More info at: http://whydoeseverythingsuck.com/2008/05/flash-10-p2p-and-cd...
It's a hybrid between BT and S3. The S3 will keep feeding the data down if there are peers or not, but if there are peers already watching, then it will utilize them as well.
I don't think this will work well for YouTube where there are millions of small videos, but for large Video distribution like NetFlix and Hulu, would be perfect.
So yeah it works if they reboot their router and restart the download but that isn't what you want to do in a youtube like app.
http://www.bbc.co.uk/blogs/bbcinternet/2008/12/introducing_i...
A public site like youtube or even Hulu wouldn't be able to make it a default client with those issues given how well Flash works, but I can see some other sites living with those prompts, especially if they offered low-res Flash and hires Bitlet.
Especially in a private account situation like Netflix, users would be more likely to accept those kind of prompts to see the videos they want in higher quality.
I'm impressed with how well it worked for me, though it's difficult to tell how much of the viewing was improved through peer-to-peer.
Good work, I think you could build something around this with the right contacts and impressive test results to show the real benefits.
Even if the CDN people don't like this, half the torrent sites out there would be better served if they had a tool like yours for quick downloads of the torrent.
I'd really focus on selling the improved user experience side, though. Ideally the end user doesn't even know that a .torrent is involved in the download. It's just another background/confusing aspect that users don't need to see.
I think companies will need a way to configure a fallback option for when BT isn't working for the user, but I think that's all just features that could eventually be worked into the mix pretty well.
(1) Do you think that this idea is new?
(2) If it is not new (and it isn't), why did previous efforts seemingly fail? Is your effort going to have those problems?
WRT Flash 10, it's tied to Adobe servers, at least for now. On the other hand, it's likely to be installed on a large fraction of the web.
I'm somewhat surprised that Silverlight doesn't (yet?) have P2P support that is server agnostic. (This would give developers/sites a reason to use silverlight instead of flash.)
I keep expecting a "not-IE" browser to include a P2P javascript extension.
If either happens, I expect that the other will happen in short order.
However, my experiences with Java have been so consistently horrible that I won't bother to try it as is. If you make a native application that I can install, and make sure the download is a few megabytes at most, I will try it. It sounds like a good idea.